Research map / Detection and segmentation / CNN features
Fast Video Object Segmentation by Reference-Guided Mask Propagation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use the reference-frame stream and previous-mask guidance together (ablation) | Siamese encoder-decoder · DAVIS-2016 validation | J Mean: 68.3 -> 81.5 | Helped |
| Use previous-mask guidance (ablation) | Siamese encoder-decoder · DAVIS-2016 validation | J Mean: 73.5 -> 81.5 | Helped |
| Pre-train on simulated samples (ablation) | Siamese encoder-decoder · DAVIS-2016 validation | J Mean: 68.6 -> 81.5 | Helped |
| Fine-tune on video data (ablation) | Siamese encoder-decoder · DAVIS-2016 validation | J Mean: 55.0 -> 81.5 | Helped |
| Train with recurrence across video frames (ablation) | Siamese encoder-decoder · DAVIS-2016 validation | J Mean: 74.3 -> 81.5 | Helped |
| Add online learning (ablation) | Ours · DAVIS-2016 validation | J Mean: 81.5 -> 82.4 | Helped |
| Add dense CRF post-processing (ablation) | Ours · DAVIS-2016 validation | F Mean: 82.0 -> 79.9 | Hurt |
| Add dense CRF post-processing (ablation) | Ours · DAVIS-2016 validation | J Mean: 81.5 -> 81.9 | Helped |
| Add a convolutional GRU visual-memory unit (ablation) | Ours · DAVIS-2016 validation | J Mean: 81.5 -> 79.6 | Hurt |
| Add a convolutional GRU visual-memory unit (ablation) | Ours · DAVIS-2016 validation | F Mean: 82.0 -> 81.0 | Hurt |
| Use the proposed method without online learning | Video object segmentation · DAVIS-2016 validation | Time: 0.13s vs 13s | Helped |
| Use the proposed method | Video object segmentation · SegTrack v2 | IoU: 77.6 -> 71.1 | Hurt |
Closest papers by meaning
- Fast Online Object Tracking and Segmentation: A Unifying Approach (CVPR 2019 · similarity 0.91)
- Siamese Network with Interactive Transformer for Video Object Segmentation (AAAI 2022 · similarity 0.90)
- Learning Dynamic Network Using a Reuse Gate Function in Semi-Supervised Video Object Segmentation (CVPR 2021 · similarity 0.90)
- From ViT Features to Training-free Video Object Segmentation via Streaming-data Mixture Models (NeurIPS 2023 · similarity 0.89)
- Per-Clip Video Object Segmentation (CVPR 2022 · similarity 0.89)
- Unified Mask Embedding and Correspondence Learning for Self-Supervised Video Segmentation (CVPR 2023 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.